Kernel: IOSurface buffer allocation size is zero

Nice one Anic264b!

Can anyone resolve anything with no knowledge? 😉

Here is a potential compromise solution.

Find a Flash movie or window on the web and right-click to open the Flash settings box. The far left tab should be a single checkbox for hardware acceleration. Uncheck it.

I've been testing a few sites without hardware acceleration and have not encountered any issues (so far). More importantly, Console is not displaying any IOSurface messages.

I imagine that any Safari resets and/or purging of Local Storage is likely to default Flash hardware acceleration to on, so just keep that in mind.

For average users like myself this works just fine, but it may not be for everyone. Users with flash heavy web-lives may well need to recheck the hardware acceleration. I would think flash gaming or animations might need it.

Anyway, it's worth seeing what it does, or doesn't do, for you.

Post your findings if you can.

A minor update to the above:

Safari resets DO NOT default the Flash settings.

Thank you all for this thread. I've lost weeks of productivity. Investigating software, then hardware to understand what might be happening with my early 2009 mbp w/ssd and NVIDIA GeForce 9400M 256 MB. I believe that there is a hardware issue that is excentuated when flash is activated. After loading 10.7 twice and having full disk encryption (which made the issue out of control) removed, I was able to track down this trend in my console that lead to my computer freezing (requiring hard reboot, but no error message upon start-up). When running some graphics, this last time a flash driven mlb.tv player, I got first the scary

kernel[0]: HFS: Very Low Disk: freeblks: 0, dangerlimit: 57

followed by 26 repetative

kernel[0]: IOSurface: buffer allocation size is zero

messages and a freeze.

I too was able to reload flash pages to generate the message, and then turned off the hardware acceleration to see that the message stopped ocurring.

I (as of two months ago) no longer have apple care and would be very grateful if someone eventually talks with 'them' about this issue and updates this post. My thinking, though limited (chemist, not computer scientist) in this area is that you've done a great job narrowing things down. So I felt compelled to corroborate what you're seeing and keep this going until it is resolved (likely by a new OS or our [urchasing of new components) or addressed by Apple.

I am doing it 6 months since I bought my Fall 2011 Mac Quad i7 2.2, now have post also in bugreport.apple.com, I am seeing they call it "elite engineer" and thats in genius, which tells me they are "bad engineers" and if you complain more you got elite 🙂 Anyway - I was there 4x times - 1 time they took my mac for 7 days and nothing and blames SSD, I returned it and put Apple HDD and still the same this time I spoke with someone and explained all the problems + plus showed them all forums, all posts, all screenshots that I have, also photo booth console IO errors and said my last step it to write to Tim Cook about either its HW - graphics or Kexts (drivers) or OS itself, because my 2007 MAC doesnt have any freeze and still going 5 years and no IO Errors.

She answered: This is over my league (genius bar) - you going to talk to "elite engineer" - I have no hope because if it is OS X bug or driver or hw - have no ability to convice them to do something cause they think all is ok....and blaming 3rd party - except they cant say anything on photo booth....will see....

Hi all! I've just spoken on apple care line, operator seemed to be suprised about problems and errors mentioned above. He suggested to put my macbook pro to genius bar, but I'm out of warranty. I told him, many people are struggling with that problems, and as people investigated over some forums, it's not hardware issue. He told me that, if 100 person report that maybe engineers will start to investigate the problem... So keep calling guys, maybe apple will fix issue which is known for more than 6 months.

I am going there and have warranty this time its my 5th turn and I am going to keep pushing them over the edge 🙂 I am web developer and I cant work so I need working mac...

rentgeen - Good luck and let me know how it works out for you. I have been battling Applecare for 7 months now on this issue and can't get them to resolve anything.

What I do know, it is NOT Flash, it IS hardware. Reading this thread clearly validates this statement as there are several different software failing yet a single hardware report in the kernel.log file. This indicates hardware causing software to fail, not the other way around. Check a normal running mac and you won't even find a kernel.log file as I have proven on 4 different mac's using the same software I am on the one that is failing.

See thread: https://discussions.apple.com/thread/3869289?answerId=18115075022#18115075022&ac_cid=142432#18115075.

In that thread you will find comments from an Apple 'Senior Advisor' insisting I need to remove 'thrid party' software because that is causing the problem. When the senior advisor was asked for comparable Apple products to software I use, CS 4, Navicat, and others; of course he didnt' bother to respond.

I too am a web developer. I have no problems at all on my iMac at work and home, but tons of issues with my mac mini...

It will be interesting to see what you find out...

Hello I havent been to Apple yet because I did crazy thing installed on the partition MOUNTAIN LION 10.8 OSX.

NO IO SURFACE ERROR IN PHOTO BOOTH!!!!

I need to try the videochat and other stuff - I am going to install also a lots of apps, cause I know LION has RAM issues...

Will post more when I find out more.

Seeing any errors in kernel is a bit scary. I need a machine, not a PC.

I am seeing these errors using Chrome. One poster noted a nice clue - there is one error per page shown in this same thread. The source of that specific error apparently is the flash-based software from Jive.com. Their software is being used by Apple's web developers. If you go to their website, using Chrome, (I am on 19.0.1084.46) you will get this error for each page or popup page used as well. The error is shown in kernel.log on the Apple console .

Jive.com is using flash player on their website. Its odd to me that a flash player can cause a kernel level error. Apple and Adobe should be working together to resolve this type of problem.

I don't agree that it is a hardware issue based upon the fact that we can reproduce using a web page. Kernel issues come from low-level OS software services that most user-level applications cannot directly bind. However, things like device drivers can be bound to the kernel. With soft kernel libraries, it is possible this is loaded at runtime - perhaps Chrome and/or Flash together create the error by simply using or calling those libraries incorrectly. Maybe Flash is doing something that passes to the kernel incorrectly. I would say Chrome and Safari are probably blameless since they aren't providing direct Apple system calls.

It could be that IOSurface isn't coded properly by allowing a user level application to generate a kernel level error. It could be not even an error - just that the OS kernel THINKS a IOSURFACE buffer size of 0 is an error and is then incorrectly reporting it. In that case, the 'freezes' could be that the kernel is simply calling out an error to its log. That kernel process is then stopping all other multi-threaded processes until the error message is sent. AFAIK, only kernel level processes have that power. So the 'freeze' is simply the lag time to push error info out to kernel.log

All that said, when I was playing with this page and basically watching the new IOSURFACE error messages show up by paging, I got an additional error message in kernel.log:

May 22 19:30:44 Bryan-Coopers-iMac kernel[0]: (default pager): [KERNEL]: ps_select_segment - send HI_WAT_ALERT

May 22 19:30:47 Bryan-Coopers-iMac kernel[0]: (default pager): [KERNEL]: Switching ON Emergency paging segment

May 22 19:30:48 Bryan-Coopers-iMac kernel[0]: (default pager): [KERNEL]: ps_vstruct_transfer_from_segment - ABORTED

May 22 19:30:48 Bryan-Coopers-iMac kernel[0]: (default pager): [KERNEL]: Failed to recover emergency paging segment

May 22 19:30:48 Bryan-Coopers-iMac kernel[0]: macx_swapon SUCCESS

there are several other kernel message around these times. So maybe it is a real error caused by Flash use of the IOSurface library. My guess is that Adobe isn't using the published IOSurface API but went lower level in use of that library.

I would say Apple and Adobe need to look further into this error as repeatable kernel level errors should be taken very seriously. Apple has created IOSURFACE to allow multiple processes to work on the same data across normal process boundaries. That sounds like Apple chose to do this at least some of this work in the kernel. Was that necessary? not sure why one would use kernel space for memory intensive inter-process work. I would also say this is potential a huge security error if a plugin for apple can directly cause kernel errors.

Apple states:

The IOSurface framework provides a framebuffer object suitable for sharing across process boundaries. It is commonly used to allow applications to move complex image decompression and draw logic into a separate process to enhance security.

Trying to dust off my aged system architect hat, maybe it has to be kernel level call since memory is usually attached to a specific process to protect other processes. However, that opens up a potential can of worms when a user level application can touch the kernel directly.

The bedrock of UNIX lies upon a stable kernel. Let user apps crash all they want- the kernel keeps us all logged in as well as secure from other applications.

Thank you for this answer! They are what we should see most in this forum (instead of “me too” answers that are meaningless). I wanted to mark your message as helpful, but it seems there's no choice for that (is that only the original poster who can do it?).

thanks for the comment!

I went over to Adobe and found they have 'closed' the associated bug 3077658:

https://bugbase.adobe.com/index.cfm?event=bug&id=3077658

They are blaming Apple OS and say that the next version of Lion 10.8, there is no error message.

Its a bit weak response in that their users are also getting full system hangs/crashes. Hence there is a real error,

but Adobe claims its all the OS and not their specific use of the library. However, when tech support closes

a memory bug, it would be nice to know there is a workaround for existing users and that the engineers reviewed their use of what they call 'third party' (Apple) code. Turning off hardware accelleration is definitely a workaround and Adobe should recommend that.

one interesting note there is that one user was getting full hangs just before a time machine sync message.

they turned off time machine, and doesn't get hangs.

that raises an interesting question - if the kernel messaging itself is part of the hang. If we have 2 kernel level processes running, and one (iosurface) wants to error out, but the other (timemachine must have some kernel level hooks to sync the file system drivers) is also messaging out, is it possible a full system hang can occur? In databaseland, we call that a livelock - when 2 processes stop waiting for another process to complete, and both are unaware of the other process. server appears hung but in fact, 2 processes are livelocked.

given history about flash and iphones, could be that adobe simply refuses to talk with apple about this problem.

The problem I think was OS because LION has also RAM issues, all dissappeared when I installed the Mountain Lion

The main problem was I think something with graphics card related - either kexts (video drivers), my mac was crashing on all video related apps: Google Video Chat, Movies, Photo Booth, and if you think about it Flash is using also video. Personally I hate flash cause I am web develeoper but in this case its Apple error. I went there 6x times and gave up and suddenly 10.8 all errors are gone. My friend has older mac and was using Snow Leopard and all was ok but he had to upgrade to Lion because of Xcode to develop iphone games and apple forces you always to upgrade. When he got to Lion his mac started crashing, then he installed 10.8 all was fine again. We tested it on 2 machines, if 10.8 fails I am going to apple again. Its not flash.

<Edited By Host>

Hi Rentgeen

It is good to hear that 10.8 doesn't show the error. It must be a combination of OSX 10.7, IOShare library, video card and user application as the IOShare library docs are pointing to a lot of use by video decompression. It makes sense that the problem has a hardware component as that proves the kernel (or at least another device driver level process) must be involved to bridge the user application space (in this case Flash Player) with another process (a video GPU) - to offload CPU time on image decompression and let the video card do that work.

A lot of guessing until I receive that source code version of OS X that I won on American Idol!

For me Lion made my work and life so hard so I am doing the research like 6 months since I bought my Fall 2011 Mac and have 5 times a day freezing and crashing. Tested 10.8 like 2 days and its Beta so I cant be sure but I was happy with the results that probably it was fixed. I have 8 GB of memory and running 6 apps and its full, tested on the 10.8 I laucnhed 38 APPS and 200 MB left. Thats huge progress. I just wanted to shw Apple something is wrong and Adobe + Apple should find a way to fix it because its pointing to 1 thing but Apple hates flash so they always will be blaming each other + I came to Apple 1st time and I said its happening on Chrome they dont even look at it and said Chrome is 3rd party and things like that....and you made a really nice point what can it be but still they wouldnt go for it, they would be looking at me like what I am talking about, my 6th time end up like - 3 geniuses told me you know more than its in our knowledge - great support! I only said you should at least made a point about this that is really something wrong with the Lion.

Does anyone experience this bug on a Mac with an ATI chipset or ATI graphics card?

In the past, all the serious problems that I have had with Macs have ALL been related to faulty/defective NVIDIA GPU chips and/or Apple's horribly buggy NVIDIA drivers.

You see Apple insists on writing their own NVIDIA drivers, instead of letting NVIDIA write them (as they do for Windows). And so the consumer (that's you and me) we have no choices over which version of the NVIDIA driver to install and use, nor can we access the settings to tweak it directly and troubleshoot these issues ourselves.

It would be fine if Apple actually fixed these problems but it's obvious they can't/won't.

Until someone convinces me otherwise then that's what I believe.

For example I have Mac Pro 2009 with NVIDIA 285 GTX. On 10.7.4 it can't even run Jedi Knight II at faster than 10 FPS. Unbelievable.

And plenty of those IOSurface errors.

I can't delete any photos off my iPad either. The OS disconnects from it.

I have to say that Lion is the worst piece of software I have ever used.